Peter's point is that sorting on a tokenized field is meaningless. Say you index "erick xu peter" and it's tokenized. You have three tokens: "erick", "xu", and "peter". What does sorting mean now? Should the document be in the e's? x's? p's?
So if you're sorting on a tokenized field, trying to understand why you get OOMs sorting desc (which I agree is kinda strange) is a waste of time. If you're NOT sorting on a tokenized field, can you answer some questions about your environment? How much memory are you giving the JVM? What version of Solr? etc. You might want to review: http://wiki.apache.org/solr/HowToContribute Best Erick On Sun, Nov 21, 2010 at 9:43 PM, xu cheng <xcheng....@gmail.com> wrote: > thanks for replying > > but when it's sort with asc, it runs pretty well > only if I sort with desc , it has the out o f memory exception > > 2010/11/17 Peter Karich <peat...@yahoo.de> > > > You are applying the sort against a (tokenized) text field? > > You should better sort against a number or a string. Probably using the > > copyField directive. > > > > Regards, > > Peter. > > > > > > hi all: > >> I configure a solr application and there is a field of type text,and > some > >> kind like this 123456, that is a string of number > >> and I wanna solr to sort the result on this field > >> however, when I use sort asc , it works perfectly ,and when I sort it > with > >> desc, the application became unacceptablly slow > >> and finally , an OutOfMemoryException was throw. > >> does anyone have the same kind of problem?or any suggestions? > >> > >> thanks > >> > >> > > > > -- > > http://jetwick.com twitter search prototype > > > > >
